|
Program Structuring in Haskell Slide 5 |
||||||||||||||||||||||||
|
||||||||||||||||||||||||
 |
|
Perhaps the most obvious candidate here is Haskell's system of type classes, which provide support for user defined overloading. In this talk, I will give some examples to illustrate how type classes can be used to promote modularity and extensibility and to reduce the coupling between different sections of a program. (However, there's a judgement call here. Some Haskell programmers have used the class mechanisms to simulate aspects of powerful module systems, like those of SML, perhaps because of weaknesses in the Haskell module system, or perhaps because the role of type classes in Haskell has been overemphasized. I consider such things to be an abuse of type classes, but it's subjective, and I won't go into it any further here.) So what is a type class? Essentially, it's just a set of types. (Think of "class" in the sense of set theory rather than object-oriented programming!) The picture here shows the set of all monotypes, represented by the kind *. Inside that is the subset of equality types, Eq, which includes things like integers and booleans, but excludes things like functions and the computable reals. What makes the Eq class special is that all of its members have something in common: they all support an effective algorithm for structural equality. |